Background. Cannabis is the most widely consumed illicit drug in the United States. Its use, particularly in early initiates, is associated with subsequent development of other drug and alcohol use disorders.
Objective. The authors examined the prevalence of cannabis use and the association between cannabis use and aberrant opioid-related behaviors in patients prescribed chronic opioid therapy for persistent pain.
Methods. PubMed was queried for studies of chronic opioid therapy in which aberrant opioid-related behaviors were quantitatively examined and in which cannabis use data (as determined by cannabinoid-positive urine drug tests) were extricable from that of other substances of abuse.
Results. The prevalence of cannabis use among patients prescribed chronic opioid therapy in these studies ranged from 6.2% to 39%, compared with 5.8% in the general United States population. Furthermore, cannabis use in chronic opioid patients shows statistically significant associations with present and future aberrant opioid-related behaviors.
Conclusion. Cannabis use is prevalent in patients prescribed chronic opioid therapy and is associated with opioid misuse. Further research is necessary to clarify the strength and the nature of the association between cannabis use and opioid misuse, and to address additional questions about the consequences of cannabis use in the context of chronic opioid therapy.
Cannabis is the third most widely consumed recreational drug in the United States after alcohol and tobacco . It is perceived by many to pose relatively little risk and, like alcohol and tobacco, is considered by many to constitute a socially acceptable lifestyle choice [1,2]. To others, however, cannabis is seen as a “gateway” to harder drugs [3,4], or even a hard drug in its own right, akin to PCP and heroin . The polarizing nature of cannabis is reflected in current law. In its natural form, cannabis is a Schedule I drug (no accepted medical use) under the federal Controlled Substance Act. More than a dozen states, however, subverting federal law, have approved the use of cannabis for medical purposes under the recommendation of a licensed physician. Δ-9-tetrahydrocannabinol (THC), the major psychoactive constituent of cannabis, has several established, and several more purported, medical benefits . Indeed, dronabinol (Marinol®, Solvay Pharmaceuticals, Brussels, Belgium), an oral synthetic THC, is a federal Schedule III drug and is Food and Drug Administration (FDA)-approved for the treatment of AIDS-related anorexia and chemotherapy-induced nausea and vomiting. More relevant to the topic of cannabis in chronic opioid therapy is an accumulating literature, much of which has demonstrated the efficacy of cannabinoids in a variety of chronic pain states . Furthermore, preclinical and recent clinical data indicate a possible enhancement of opioid analgesia by cannabinoids [8–10]. Sativex® (GW Pharmaceuticals, Vashon, WA), a cannabinoid product derived from the cannabis plant, has been approved in Canada for neuropathic pain associated with multiple sclerosis and for opioid-resistant cancer pain. It is currently in phase IIB clinical trials for cancer pain in the United States, with phase III trials expected to begin in 2010 (Ethan Russo, GW Pharmaceuticals, personal communication, June 22, 2009). Given the controversial nature of cannabis—particularly the smoked botanical product—it might be expected that physicians differ in the significance they place on patients' use of this drug in the context of chronic opioid therapy for nonmalignant pain [11–13].
A previous or current history of substance abuse is an established predictor of opioid misuse among patients prescribed chronic opioid therapy . Cannabis use, however, has not yet been systematically examined as a risk factor for aberrant opioid-related behaviors. The aim of this study was to examine the prevalence and significance of smoked cannabis independently from other drugs of abuse in patients prescribed chronic opioid therapy.
In order to identify studies in which cannabis use in chronic opioid therapy was addressed, the authors queried the PubMed database using the following combinations of search terms: marijuana OR cannabis AND opioids AND chronic pain; marijuana OR cannabis AND urine toxicology AND chronic pain; marijuana OR cannabis AND urine drug AND chronic pain; opioids AND chronic pain AND urine toxicology OR urine drug; opioid therapy AND chronic pain AND abuse OR misuse OR aberrant. Searches were limited to English language journals and human subjects. Relevant studies were also identified through bibliographies of retrieved manuscripts and from studies available in the authors' files. On the basis of titles and abstracts, complete manuscripts were reviewed by the authors.
Studies were eligible for inclusion if they included data on both the prevalence of cannabis use among persons with chronic pain and prevalence of aberrant opioid-related behaviors. Studies were excluded if cannabis use was not confirmed by means of urine drug testing or if cannabis data were not extricable from that of other substances of abuse. The intent of this review was not to conduct a systematic meta-analysis. Rather, it was to identify studies that include data on use of cannabis among pain patients prescribed chronic opioid therapy, regardless of study design, practice setting, patient demographics, definitions of aberrant opioid-related behaviors, and/or indications for urine drug testing.
Altogether, 2,237 titles were identified. The search was reduced to 43 separate titles when identifying clinical studies related to cannabis. Of these, 14 met inclusion criteria for review. The studies were reviewed separately regarding the prevalence of cannabis use among patients prescribed chronic opioid therapy in specialty pain clinics and primary care settings.
Specialty Pain Clinics
Several studies from specialty pain centers, nearly all of which have been published within the past 5 years, have included data on the prevalence of cannabinoid-positive urine drug screens among patients prescribed chronic opioid therapy. Katz et al. , in a retrospective study of two New England academic pain clinics (N = 122; mean age 45; range 22–82), one of which performed urine drug testing at each visit and one of which performed testing annually and for-cause, reported cannabinoid-positive urines in 16.4% of patients. Wasan et al. , in a prospective study of patients prescribed chronic opioid therapy in academic and private anesthesiology-based pain clinics in Boston, MA, Lehigh Valley PA, and Northwest, IN (N = 228; mean age 50; range 21–89), found that 31.8% of urine drug screens were cannabinoid-positive. Patients were stratified into high- and low-psychiatric comorbidity groups based on responses to the psychiatric subscale of the Prescription Drug Use Questionnaire. Reanalysis of the raw study data demonstrated that cannabinoid-positive urine drug screens were more likely in the high-psychiatric comorbidity group (41%) than in the low-psychiatric comorbidity group (26%) (X2 = 4.7; P = 0.03).
Michna et al. , in a retrospective review of random urine drug test results in a Boston academic pain clinic (N = 470; mean age 47; range 21–85), found cannabinoid-positive urines in 14.6% of patients. An earlier study by the same group, in which patients were subjected to for-cause urine drug testing (N = 61; mean age and age range unspecified), found that 21% of patients yielded cannabinoid-positive urines .
Manchikanti et al., in the setting of a private interventional pain center in Kentucky, have published four pertinent prospective studies in recent years. In the most recent work , 11% of 500 consecutive patients (mean age 49; range 21–78) undergoing interventional pain management procedures, all of whom were prescribed stable doses of opioids, produced cannabinoid-positive urines. In an earlier work , 200 patients—100 consecutive patients receiving hydrocodone (mean age 49; range 22–75) and 100 consecutive patients receiving methadone (mean age 48; range 22–75)—were subjected to urine drug testing. No patients were suspected of opioid misuse. Eighteen percent produced cannabinoid-positive tests (16% in the hydrocodone group and 20% in the methadone group). In a 2004 study , 200 patients prescribed chronic opioid therapy—100 consecutive patients seeking additional opioids (mean age 49; range 23–76) and 100 consecutive patients not seeking additional opioids (mean age 46; range 21–85)—were subjected to urine drug testing; 20% of patients produced cannabinoid-positive tests (25% in the group seeking additional opioids and 15% in the group not seeking additional opioids). In a 2003 study , 100 consecutive patients without a history of drug abuse, undergoing interventional pain management, and receiving chronic opioid therapy (ages unspecified) were randomly selected for urine drug testing. Cannabinoids were detected in the urine of 14% of patients.
Atluri et al. , in two private interventional pain management centers in the Cincinnati area, performed urine drug testing in patients prescribed chronic opioid therapy (N and ages unspecified), randomly and for-cause. Of note, all patients were asked about marijuana use prior to drug testing, and only patients who denied recent cannabis use were tested. Nevertheless, 39% of patients produced cannabinoid-positive urine drug tests.
Fishbain et al.  prospectively evaluated consecutive patients (N = 274; mean age 46) admitted to a pain facility in Miami, FL. Each patient was asked to consent to urine drug testing; 6.2% produced cannabinoid-positive specimens. Of note, 13.9% refused urine drug testing and an additional 3.6% had test results that could not be located by the authors.
Most recently, Cone et al. , tested 13,948 urine specimens collected during 2006 from 31 pain clinics in Tennessee, West Virginia, Kentucky, Ohio, Florida, and Indiana—none of which have medical marijuana laws. All specimens were screened by immunoassay for amphetamines, barbiturates, benzodiazepines, cannabinoids, carisoprodol, cocaine, ecstasy, fentanyl, meperidine, methadone, opiates, oxycodone, and propoxyphene, and positive screens were confirmed by gas chromatography-mass spectrometry. The cannabis metabolite 11-nor-9-carboxy-Δ9-tetrahydrocannabinol was detected in 8.9% of the 10,992 specimens that confirmed positive for at least one drug or metabolite. The authors did not report patient demographic information.
Primary Care Clinics
Similar results have been found in primary care clinics. Hariharan et al.  performed a 5-year retrospective study in their academic general internal medicine clinic in Milwaukee, WI. Patients receiving chronic opioid therapy for nonmalignant pain (N = 300; median age 49) were subjected to urine drug screens. The methodology stipulated that “random urine drug screening would be performed if recommended by the physician to monitor adherence and possible use of illicit substances.” There was no randomization protocol, however, and only 45% of patients were subjected to urine drug testing (UDT). The authors speculated that UDT may have been selectively performed on those patients suspected of being at elevated risk of substance abuse. Twenty percent of patients produced cannabinoid-positive results.
Fleming et al.  performed a study encompassing 235 primary care physicians throughout eight counties in Wisconsin. Patients prescribed chronic opioid therapy for nonmalignant pain (N = 801; mean age 49; range 18–81) were voluntarily recruited for a pain and substance use interview, followed by a urine drug test. Cannabinoid-positive results were found in 20.2% of patients.
Ives et al.  prospectively evaluated the 1-year incidence and prevalence of opioid misuse among patients (N = 196; mean age 52; range 26–85) in an academic primary care referral pain clinic. The clinic encouraged referrals of patients with management difficulties and suspected opioid misuse. Eighteen percent of patients produced cannabinoid-positive urines during the study period.
Significance of Cannabis Use among Patients Prescribed Chronic Opioid Therapy
The significance of the data from pain and primary care clinics becomes apparent when given context. Table 1 provides data from the National Survey on Drug Use and Health (NSDUH) on the prevalence of smoked cannabis use across age groups in the general United States population for the past several years. We were able to retrieve and reanalyze the raw data from the single center study by Michna et al.  and the multicenter study by Wasan et al. , and found that percentages of current cannabis users were higher—usually by a multiple—than those in nearly every age group in the general population as determined by NSDUH (Table 2). Comparisons were run between the NSDUH data and the Michna et al. (2007) and Wasan et al. (2007) data. Comparative results from these studies demonstrated statistically significant differences between the general population and chronic opioid populations (X2 = 19.34; df = 2; P = 0.001).
Source: SAMHSA, Office of Applied Studies, National Survey on Drug Use and Health, 2002–2007.
|Michna et al. (2007)||Wasan et al. (2007)*|
|Age Group||# in Group||# THCA-Positive||% THCA-Positive||# in Group||# THCA-Positive||% THCA-Positive|
|Michna et al. (2007)||Wasan et al. (2007)*|
|Age Group||# in Group||# THCA-Positive||% THCA-Positive||# in Group||# THCA-Positive||% THCA-Positive|
Data available for 193 of 228 subjects.
THCA = Δ-9-tetrahydrocannabinolic acid.
Available evidence indicates that cannabis use in patients prescribed chronic opioid therapy is associated with current and future opioid misuse. For example, Manchikanti et al.  found that 13.8% of patients with cannabinoid-positive urines were current opioid abusers (defined as engaging in doctor shopping or legally adjudicated opioid trafficking), compared with 9.2% of all 500 patients, a statistically significant difference (data derived from L. Manchikanti, personal communication, July 14, 2008).
Ives et al. , in their previously noted study, found that one or more cannabinoid-positive urine drug tests was a strong predictor of next 12-month opioid misuse (33% vs 12%; P = 0.001), with opioid misuse defined as one or more of the following: 1) two or more instances of urine drug screens negative for a prescribed opioid or other controlled substance; 2) two or more instances of urine drug screens positive for a nonprescribed opioid or other controlled substance; 3) evidence of procurement of opioids or other controlled substances from multiple providers; 4) evidence of diversion of prescribed opioids or other controlled substances; 5) evidence of prescription forgery; and 6) confirmed positive urine drug tests for cocaine or amphetamine.
Furthermore, several studies show strong associations between cannabis and opioid abuse among individuals seeking treatment for opioid-use disorders [31–35]. For example, one recent study found that, among individuals entering an opioid detoxification program, cannabis was the most frequently co-abused drug more frequently abused than heroin, cocaine, and amphetamines .
Fleming et al. , in their previously noted study, found cannabinoid-positive drug screens to be predictors of any current substance use disorder (OR 3.52; P < 0.001). This group recently reanalyzed data from their 2007 study . Analysis of patients with cannabinoid-positive urine results revealed that, among patients who displayed 0/12 aberrant (mostly opioid-related) behaviors, 13.6% produced cannabinoid-positive urines; among patients who displayed 1–3/12 aberrant behaviors, 15.9% produced cannabinoid-positive urines; and among patients who displayed ≥4/12 aberrant behaviors, 28.0% produced cannabinoid-positive urines. Logistic regression predicted ≥4/12 aberrant behaviors on the basis of cannabinoid-positive urine drug test results (OR 1.52, CI 1.03–2.23, P = 0.034). Of note, the frequency of any substance use disorder was 0.6% in the group with 0/12 aberrant behaviors, and 24.2% in the group with ≥4/12 aberrant behaviors.
Finally, Ecklund and Ellias reported their experience in a letter to the editor . The authors prospectively followed all patients in their pain clinic (N not specified) who had produced cannabinoid-positive urine drug tests. At 6 months, each of these patients had committed an opioid contract violation (including use of illicit and unprescribed licit drugs, altering prescriptions, and drug trafficking).
A modest existing literature indicates 1) a high prevalence of cannabis use among patients prescribed opioid therapy for chronic nonmalignant pain, and 2) an association between cannabis use and current and future aberrant opioid-related behaviors. These associations exist in academic and private practice settings, specialized pain clinics and primary care clinics, and urban and nonurban practices. Furthermore, cannabis does not appear to be merely an epiphenomenon of polysubstance or hard drug abuse. In the five studies in which cannabis was unbundled from other drugs of abuse, cannabis was the only drug of abuse detected in 71–93% of cases [18,20–22,25].
This association between cannabis use and opioid misuse has also been demonstrated outside of the pain literature. Recent analysis of data from the 2003 NSDUH demonstrated an association between past year cannabis use and past year nonmedical use of prescription opioids (AOR 2.03, 95% CI 1.61–2.55 for women; AOR 2.24, 95% CI 1.74–2.88 for men) . More recent analysis of data from the 2005 and 2006 NSDUH demonstrated a strong association between past year cannabis use and past year nonmedical use of prescription opioids in persons aged 50 and older. The authors reported a past-year prevalence rate of nonmedical opioid use of 10.7% in persons acknowledging past-year cannabis use, compared with a rate of 1.1% in persons not acknowledging past-year cannabis use (AOR 7.07, 95% CI 3.99–12.53) .
There are several nonmutually exclusive hypotheses that may explain the association between cannabis and opioid abuse . First, there may be a noncausal association due to common exposure opportunity, i.e., cannabis sources are also sources of initial nonmedical opioid exposure. Second, there may be a noncausal association due to common risk factors, e.g., genes for novelty-seeking or unhealthy peer affiliations. Third, this patient population may be self-medicating with cannabis for their chronic pain and/or psychiatric comorbidities. And fourth, there may be a causal association whereby neurophysiologic effects of early cannabis exposure increase vulnerability to the reinforcing effects of subsequent opioid exposure. Converging lines of laboratory evidence demonstrating significant functional interactions between the endogenous cannabinoid and opioid systems lend support to this hypothesis. For example, in the rat: 1) Cannabinoid (CB1) and opioid (MOR) receptors are co-localized in the nucleus accumbens (NAc) shell and ventral tegmental area (VTA), key anatomical substrates for reward . 2) Both THC and opioids have been shown to selectively stimulate dopamine transmission in the NAc shell. THC appears to effect this stimulation through the MOR1 receptors in the VTA, suggesting that it may play a role—through pharmacological “priming”—in subsequent opioid abuse . 3) Chronic administration of THC induces cross-tolerance to opioids . Adults exposed to THC during adolescence display an increased susceptibility to opioid self-administration. 4) This behavior is associated at the molecular level with altered MOR1 function and discrete alterations of endogenous opioid activity in the VTA and NAc shell .
Notwithstanding the actual and potential medical benefits of cannabinoids, the association of smoked cannabis with opioid misuse is concerning. The strength and nature of this association should be addressed in future studies of aberrant drug-related behaviors in the context of chronic opioid therapy. It is important to unbundle cannabis data from that of other drugs of abuse and, where possible, to stratify cannabinoid-positive urine drug screens according to patient demographic data such as age and gender. Our observations suggest several research questions deserving of investigation. For example, what are the specific types of aberrant opioid-related behaviors displayed by cannabis users, and how, if at all, do these behaviors differ from those seen in patients who abuse other drugs? What is the age of onset of cannabis use in these patients? And is early initiation of cannabis use predictive of subsequent opioid misuse in chronic pain patients? It has been established that early initiation of cannabis use is a significant risk factor for subsequent development of other drug and alcohol use disorders . Does cannabis use diminish pain in all patients ? If so, do cannabis-using patients require less opioid? Does cannabis adversely affect functional outcomes in pain patients on chronic opioid therapy? If so, does cessation of cannabis use improve these outcomes? Does cannabis use represent occasional use, abuse, or dependence? And what is the purpose of cannabis use in these patients—recreational or “medicinal”? Unfortunately, answers to the latter two questions may be elusive because users of illicit drugs tend to be dishonest about their drug use.
The results of this review suggest other avenues for investigation. For example, acute cannabis use increases the risk for drugged driving crashes [42–44]. Residual psychomotor effects of the drug may last 24 hours or longer [45–47]. What then is the effect of cannabis combined with opioids (and other commonly used drugs in pain management, including benzodiazepines)? And what are the legal consequences of prescribing chronic opioid therapy to patients known to be cannabis users? The U.S. Drug Enforcement Administration (DEA) position on cannabis is unequivocal: until the FDA approves its use for a particular disease or condition, cannabis is not a legitimate medication. The DEA identifies it as a precursor to abuse of other drugs; a factor in drugged driving; and a danger to the user and to others .
There are a number of limitations of any review of the literature of this nature. First, it is possible that some studies were omitted from this review either because they were non-English or the search process was not complete. In some articles, the description of the urine screening methods and results lacked detail. Some studies self-selected those who would be screened for cannabis, possibly inflating the percent of positive results. Finally, cannabis use in these studies was determined by urine drug screens, whereas NSDUH data were derived from anonymous self-report. None of these limitations, however, weaken the association between cannabis use and current and future opioid misuse.
Cannabis is the most widely consumed illicit recreational drug in the United States. Its use is several times more prevalent in pain patients who are prescribed chronic opioid therapy than in the general population. As well, its use in this population is associated with present and future aberrant opioid-related behaviors. Further study is needed to confirm these findings, to clarify the nature of the association between cannabis use and opioid misuse, and to address additional questions about the consequences of cannabis use in chronic opioid therapy.
This study was supported in part by grants (R21 DA024298, Jamison, PI; K23 DA020682, Wasan, PI) from the National Institute on Drug Abuse (NIDA) of the National Institutes of Health, Bethesda, MD, and the Arthritis Foundation (Investigator Award; Wasan, PI). The authors would like to especially thank Stephen Butler for his assistance. There are no conflicts of interest to declare associated with this study.